DOP070Endoscopic phenotypes predict outcomes and align with transcriptomic profiles in immune checkpoint inhibitor–induced colitis: A multicentre cohort study

Abstract Background Immune checkpoint inhibitors (ICI) have transformed cancer outcomes but often cause ICI- colitis (ICI-c) leading to ICI discontinuation, morbidity, and prolonged immunosuppression. Published cohorts are limited by heterogeneous definitions and modest sample size while prospective trials are scarce. We report the largest rigorously defined ICI-c cohort, including patients (pts) with endoscopic +/- histologic inflammation following ICI- diarrhoea. Methods Retrospective data from consecutive ICI-c pts at 2 UK centres were analysed. Endoscopic severity was defined by MES; endpoints included week-12 corticosteroid-free remission (CFCR). Associations were analysed using chi-squared tests and multivariable logistic regression. RNA sequencing of colonic biopsies from 18 ICI-c pts and 18 healthy controls enabled gene- and pathway-level analyses. Results Of 259 pts (53% male, median age 63) cancers included melanoma (43%), renal (19%), lung (16%) and others. ICI therapies included anti-PD-1/L1 (38%), dual anti-CTLA-4/PD-1 (44%), chemo-ICI (8%) and others. Endoscopy showed MES 0 in 33% (n = 86), MES 1 in 47.5% (n = 123), MES 2 in 12% (n = 31), and MES 3 in 7.3% (n = 19). 56% required hospitalisation (median 15 days), which correlated with CTCAE grade (p 0.001) but not MES. CTCAE diarrhoea grade did not correlate with MES, whereas bloody stool (19%) did (p 0.0001). Faecal calprotectin (FC) trended higher with increasing MES (p = 0.06) but did not vary by CTCAE grade; CRP did not correlate with MES. Concurrent CMV (n = 20) was was linked to higher MES and steroid (cs) refractoriness. ICI-microscopic colitis was more common with anti-PD-1 (76% vs 51%, p = 0.026) and associated with longer cs therapy (119 vs 68 days, p = 0.013), and longer diarrhoea duration (62 vs 46 days, p = 0.15). Male sex (OR 2.5, p = 0.002) and dual ICI (OR 3, p = 0.001) predicted macroscopic inflammation (MES 1-3), while renal cancer was protective (OR 0.38, p = 0.018). Macroscopic inflammation was linked to higher FC (771 vs 350, p = 0.02), greater need for advanced therapy (OR 2.04, p = 0.009), and higher recurrence after CFCR (OR 5.5, p 0.0001). MES 0 ICI-c (n = 6) exhibited 793 DEGs, most overlapping with the 3,771 DEGs in MES 1–3 pts (n = 12), indicating shared transcriptional alterations. Shared pathways included interferon and TNF signalling, with MES 1–3 showing amplified inflammatory responses and additional pathways such as TGFβ signalling and unfolded protein response. Conclusion In this large real-world CPI-colitis cohort, macroscopic inflammation predicted a more aggressive course, with higher FC and greater need for advanced therapy. Endoscopic phenotypes identified clinically meaningful subgroups, and transcriptional profiling revealed a graded inflammatory landscape reflecting MES. Conflict of interest: Ibraheim, Hajir: No conflict of interest D’arienzo, Paolo: No conflict of interest Dean, Nathan: No conflict of interest Manickavasagar, Thubeena: No conflict of interest Aboulela, Mohamed: No conflict of interest Oommen, Irene: No conflict of interest Mullings, Stephanie: No conflict of interest Anderson, Kirsty-ellen: No conflict of interest Vadera, Shree: No conflict of interest Singh, Ashish: No conflict of interest Kalofonou, Foteini: No conflict of interest Symington, Jake: No conflict of interest Dar, Sabeera: No conflict of interest Campbell, Michael: No conflict of interest Muehlschlegel, Paula: No conflict of interest Olsson-Brown, Anna: Roche Pharma, Eli Lilly, UCB Pharma, Novartis, BMS Young, Kate: No conflict of interest Powell, Nick: Grant: Takeda, BMS, Pfizer, Astra-Zeneca Personal Fees: Abbvie, Abivax, Allergan, Astra-Zeneca, Bristol-Myers Squibb, Celgene, Celltrion, Dr Falk Pharma UK Ltd, Ferring, Galapagos, GSK, Janssen, MSD, Roche, Pfizer, Sobi, Takeda, Tillotts